za.co.absa.cobrix.cobol.parser.exceptions.SyntaxErrorException: Syntax error in the copybook at line 95: Invalid input '999.9999' at position 95:63

[BharathBejoy]

Background

I'm using za.co.absa.cobrix:spark-cobol_2.12:2.7.9 Maven Library in Azure Databricks to load Cobol (EBCDIC) files into Databricks. It works very well for all files except one.

Error Message

Py4JJavaError: An error occurred while calling o462.load.
: za.co.absa.cobrix.cobol.parser.exceptions.SyntaxErrorException: Syntax error in the copybook at line 95: Invalid input '999.9999' at position 95:63

CPY File (Copybook)

Question

I'd appreciate some assistance in resolving this issue.

[yruslan]

Hi, thanks for the issue report! I'll try to reproduce it.

However, 'VALUE' clauses are ignored by Cobrix. It never validates that the corresponding column has one of the listed values. So you can safely comment or delete line 95 to make it work.

For the parser check, could I ask you to send me lines 92-96 of the copybook? I see '1942' there and can't put it into the context.

[BharathBejoy]

Hi, thanks for the issue report! I'll try to reproduce it.

However, 'VALUE' clauses are ignored by Cobrix. It never validates that the corresponding column has one of the listed values. So you can safely comment or delete line 95 to make it work.

For the parser check, could I ask you to send me lines 92-96 of the copybook? I see '1942' there and can't put it into the context.

Hi @yruslan, Please have a look at the attached image for the lines you asked for from the copybook.

[yruslan]

I confirm this is a parser issue. It doesn't properly handle dot as part of the number. As a workaround you can replace +999.9999 with +9999999 or '+999.9999'. Let's keep this issue open, will fix the issue if it is not too hard.

[yruslan]

The grammar has support for dots in numeric literals:

cobrix/cobol-parser/src/main/scala/za/co/absa/cobrix/cobol/parser/antlr/copybookLexer.g4

Line 229 in ae3fedf

NUMERICLITERAL: [0-9]* DOT? [0-9]+ (E SIGN_CHAR? [0-9]+)?;

I can reproduce the issue, but it is not obvious how to fix it. So will keep it open for now.

[BharathBejoy]

Hi @yruslan, DOT in numeric literals is causing this issue. Will replacing +999.9999 with +9999999 or '+999.9999' in copybook affect or change the output data in any way? Please advise. Thank you!

[yruslan]

Not at all. Changing that literal won't affect data in any way. Cobrix ignores VALUE clause.

It is an interesting issue. If the literal contains any digit but 9, it passes. So you can replace +999.9999 with +999.99990 safely, or to any other number.

[yruslan]

Hi @tr11 , I hope you are doing well! This issue might be interesting to you too